A deep soil sampling method based on geological exploration

Abstract

The present invention discloses a deep soil sampling method based on geological exploration, and relates to the technical field of geological exploration. The method comprises the following steps: obtaining a dimensional sequence of soil information of each depth layer from a geological exploration area, wherein each depth layer corresponds to a soil parameter value; and obtaining an influence factor of each depth layer on soil characteristics based on the difference between the soil parameters of each depth layer. The present invention can effectively obtain soil information of different depth layers, and by calculating the influence factor and importance of each depth layer on soil characteristics, adjusting the initial clustering distance, and finally obtaining the optimal clustering distance, reasonably setting the neighborhood radius and the minimum number of samples, and using a density clustering algorithm to implement clustering operations of different depth layers, thereby constructing a soil characteristic prediction model, accurately predicting soil characteristics of different depth layers, effectively reducing manual operations and time costs, and improving the accuracy and reliability of soil characteristic prediction.

Description

Technical Field

[0001] The invention relates to the technical field of geological exploration, and in particular to a deep soil sampling method based on geological exploration. Background Art

[0002] "Geological exploration" refers to the investigation and research activities that use various means and methods to survey and detect geology, determine the appropriate bearing layer, determine the foundation type according to the bearing capacity of the bearing layer, and calculate the foundation parameters. It is the investigation and research work on the geological conditions such as rocks, strata, structures, minerals, hydrology, landforms, etc. in a certain area to find out the quality and quantity of minerals, as well as the technical conditions for mining and utilization, and provide the mineral reserves and geological data needed for mine construction design.

[0003] In recent years, with the development of data processing and analysis technology, intelligent detection and analysis methods based on geological information have gradually received attention. However, when dealing with deep soil characteristics analysis, existing technologies face problems such as high data dimensions but sparse samples, and uneven influence of soil layers on soil characteristics. It is difficult to fully and accurately reflect the soil characteristics of different depths, thus affecting the subsequent geological application effect. Summary of the invention

[0004] The purpose of the present invention is to provide a deep soil sampling method based on geological exploration to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a deep soil sampling method based on geological exploration, the method comprising the following steps:

[0006] From the geological exploration area, obtain a dimensional sequence of soil quality information for each depth layer, wherein each depth layer corresponds to a soil quality parameter value;

[0007] Based on the differences between soil parameters at each depth layer, the influencing factors of each depth layer on soil properties are obtained;

[0008] Based on the differences among all depth layers and the influencing factors of each depth layer on soil properties, the weight of each depth layer on soil properties is obtained;

[0009] Based on the weight of each depth layer on soil properties, the initial clustering distance between different depth layers is adjusted to obtain the optimal clustering distance;

[0010] Based on the priority clustering distance, the neighborhood radius and the minimum number of samples are obtained respectively, and the density clustering algorithm is used to perform clustering operations on different depth layers to obtain the clustering results;

[0011] Based on the clustering results, a soil property prediction model is constructed, and the soil properties of different depth layers are predicted using the model.

[0012] As a specific solution of the technical solution of this application, the influencing factor of each depth layer on soil properties includes the following steps:

[0013] Record each depth layer as a data point;

[0014] Among all the depth layers, any one depth layer is recorded as the target layer, and the other depth layers except the target layer are recorded as reference layers;

[0015] Use the elbow method to operate on the i-th reference layer to obtain the optimal number of clusters for the i-th reference layer;

[0016] Based on the optimal number of clusters in the i-th reference layer, the K-means clustering algorithm is used to cluster all data values ​​in the i-th reference layer to obtain several clusters;

[0017] When the data values ​​in all reference layers corresponding to the ith data point are in the clusters corresponding to all references in the ith cluster combination, the ith data point is recorded as the mapping data point of the ith cluster combination.

[0018] As a specific solution of the technical solution of the present application, a plurality of main mapping data points are selected from all the mapping data points based on each cluster combination;

[0019] According to the differences between the soil parameter values ​​corresponding to the main mapping data points of each cluster combination, the influencing factors of the target layer on the soil characteristics are obtained.

[0020] As a specific solution of the technical solution of this application, the importance of each depth layer to soil properties includes the following steps:

[0021] The soil parameter value of the target layer corresponding to the data point is taken as the horizontal axis, and the soil parameter value of the i-th reference layer corresponding to the data point is taken as the vertical axis, and a three-dimensional data space of the target layer and the i-th reference layer corresponding to all data points is constructed;

[0022] In the three-dimensional data space of the target layer and each reference layer, a two-dimensional data space is constructed with all data points having the same horizontal coordinate value;

[0023] Based on the influencing factor of each depth layer on soil properties and the number of data points in the two-dimensional data space divided by the three-dimensional data space of the target layer and each reference layer, the importance of the target layer to the soil properties is obtained.

[0024] As a specific solution of the technical solution of the present application, the determination of the preferred clustering distance includes the following steps:

[0025] With the soil parameter value of each depth layer corresponding to the data point as the vertical axis and the soil characteristic value corresponding to the data point as the horizontal axis, a plane coordinate system for each depth layer corresponding to all data points is constructed;

[0026] Based on the importance of each depth layer to soil properties and the initial cluster distance of any two data points on the plane coordinate system of each depth layer, the cluster distance between any two data points is obtained.

[0027] As a specific solution of the technical solution of the present application, the preferred clustering distance determines the neighborhood radius and the minimum number of samples, including the following steps:

[0028] The mean of the preferred clustering distances between any two data points is recorded as the neighborhood radius;

[0029] All other data points whose preferred clustering distance to the nth data point is less than the neighborhood radius are recorded as neighborhood data points of the nth data point;

[0030] Count the number of neighborhood data points of each data point, round up the mean of the number of neighborhood data points of all data points, and record it as the minimum sample number.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] This deep soil sampling method based on geological exploration can effectively obtain soil information of different depth layers. By calculating the influencing factors and importance of each depth layer on soil properties, adjusting the initial clustering distance, and finally obtaining the optimal clustering distance, the neighborhood radius and the minimum number of samples are reasonably set, and the density clustering algorithm is used to realize clustering operations of different depth layers, thereby constructing a soil property prediction model to accurately predict the soil properties of different depth layers, which can effectively reduce manual operations and time costs, and can also improve the accuracy and reliability of soil property prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A schematic diagram of the process of the deep soil sampling method for geological exploration of the present invention;

[0034] Figure 2 A schematic diagram of the process flow of factors affecting soil properties at each depth layer of the present invention;

[0035] Figure 3 A schematic diagram of the importance of each depth layer to the soil quality graph of the present invention;

[0036] Figure 4 A schematic diagram of the process of determining the domain radius and the minimum number of samples for the preferred clustering distance of the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] It should be noted that, in the description of the present invention, the terms "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" etc. indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as a limitation on the present invention.

[0039] Furthermore, it should be understood that for the sake of ease of description, the sizes of the various components shown in the drawings are not drawn according to actual proportions. For example, the thickness or width of certain layers may be exaggerated relative to other layers.

[0040] It should be noted that like reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined or described in one drawing, it will not require further detailed discussion and description in the description of the subsequent drawings.

[0041] like Figure 1-Figure 3 As shown, the present invention provides a technical solution: a deep soil sampling method based on geological exploration, comprising the following method, obtaining a soil information dimension sequence of each depth layer from a geological exploration area, wherein each depth layer corresponds to a soil parameter value;

[0042] Based on the differences between soil parameters at each depth layer, the influencing factors of each depth layer on soil properties are obtained;

[0043] Based on the differences among all depth layers and the influencing factors of each depth layer on soil properties, the weight of each depth layer on soil properties is obtained;

[0044] Based on the weight of each depth layer on soil properties, the initial clustering distance between different depth layers is adjusted to obtain the optimal clustering distance;

[0045] Based on the priority clustering distance, the neighborhood radius and the minimum number of samples are obtained respectively, and the density clustering algorithm is used to perform clustering operations on different depth layers to obtain the clustering results;

[0046] Based on the clustering results, a soil property prediction model is constructed, and the soil properties of different depth layers are predicted using the model.

[0047] It should be clear that in the embodiment of the present application, in order to be able to obtain 5 depth layers in the geological exploration area, they are recorded as D1, D2, D3, D4, and D5. Each depth layer corresponds to a soil parameter value, such as clay content, silt content, dry volume, organic matter content, and total salt content. Through geological exploration and soil analysis, the following soil information dimension sequence can be obtained:

[0048] Depth Layer Clay content% Powder content% Dry capacity g / cm³ Organic matter content (%) Total salt content (g / kg) D1 15 45 1.4 2.5 5 D2 20 40 1.5 2.8 4.5 D3 18 42 1.45 2.6 5.5 D4 22 38 1.6 3.0 4.8 D5 16 44 1.48 2.7 5.2

[0049] Based on the difference between the soil parameters of each depth layer, the absolute deviation or Euclidean distance is used to calculate the impact factor of each depth layer on the soil characteristics. For example, the Euclidean distance is used to calculate the difference between D1 and D2: Euclidean distance = √((15-20)²+(45-40)²+(1.4-1.5)²+(2.5-2.8)²+(5-4.5)²) = √(25+25+0.01+0.09+0.25)≈ √50.35 ≈ 7.096.

[0050] Based on the differences between all depth layers and the influencing factors of each depth layer on soil properties, the weighted average method can be used to calculate the weight of each depth layer on soil properties. The weight can be determined according to the relative size of the influencing factors. The larger the influencing factor, the higher the weight.

[0051] In order to obtain the influencing factor of each depth layer on soil properties, each depth layer is recorded as a data point. Each depth layer is a two-dimensional data point composed of clay content and organic content. Among all depth layers, any depth layer is recorded as the target layer, and other depth layers except the initial target layer are recorded as reference layers. If D1 is selected as the target layer, then D2, D3, D4, and D5 are reference layers. The elbow method is used to calculate the i-th reference layer to obtain the optimal number of clusters for the i-th reference layer. The optimal K value is found by calculating the SSE under different K values. The smaller the SE, the closer the data points are to their cluster centers. For each reference layer, the SSE is calculated when K=1, 2.n, and a graph of the relationship between SSE and K is plotted. The optimal K value is determined by observing the elbow position in the graph. After calculation, it is found that K=2 is a reasonable choice for all reference layers. Based on the optimal number of clusters in the i-th reference layer, the K-means clustering algorithm is used to cluster all data values ​​in the i-th reference layer to obtain several clusters. The K-means algorithm (K=2) is used to cluster D2, D3, D4, and D5. As the clustering results are as follows, cluster 1: the soil parameters of the two depth layers D2 and D5 are relatively close, and cluster 2: the soil parameters of the two depth layers D3 and D4 are relatively close. When the data values ​​in all reference layers corresponding to the ith data point are in the clusters corresponding to all references in the ith cluster combination, the ith data point is recorded as the mapping data point of the ith cluster combination, and the relationship between the target layer D1 and the reference layer clustering results is checked. By calculating the distance between D1 and the center of cluster 1 and the center of cluster 2, it is found that the soil parameters of D1 are closer to cluster 1. Therefore, D1 is recorded as the mapping data point of cluster 1.

[0052] The method comprises screening out a number of main mapping data points from all the mapping data points based on each cluster combination; in the present application, the main mapping data point refers to a data point in each cluster combination that can represent the overall characteristics or significant characteristics of the cluster, and this data point can be the center point, boundary point and extreme point of the cluster.

[0053] According to the difference between the soil parameter values ​​corresponding to the main mapping data points of each cluster combination, the influencing factor of the target layer on the soil properties is obtained. Two clusters are obtained by the K-means clustering algorithm, which are recorded as Cluster 1 and Cluster 2. In each cluster, we can select the center point as the main mapping data point. The main mapping data point of Cluster 1 can be the mean point of all data points in the cluster, recorded as Mean_Cluster_1, and the main mapping data point of Cluster 2 can also be the mean point of all data points in the cluster, recorded as Mean_Cluster_2. For each target layer data point, the difference in soil parameter values ​​between it and the main mapping data point of each cluster combination is calculated. The smaller the distance between the target layer data point and the main mapping data point of a cluster combination, the more similar the soil properties of the target layer data point and the cluster are. Therefore, it can be considered that the soil properties of the cluster have a greater impact on the target layer. According to the distance between the target layer data point and the main mapping data point of each cluster combination, a weight is assigned to each cluster. The larger the weight, the greater the influence of the soil characteristics of the cluster on the target layer. Assume that the target layer data point is Target_Point, and its soil parameter value is (X_Target, Y_Target). We can calculate the difference in soil parameter values ​​between Target_Point and Mean_Cluster_1 and Mean_Cluster_2, recorded as Dist_Cluster_1 and Dist_Cluster_2. If Dist_Cluster_1 < Dist_Cluster_2, it can be considered that the soil characteristics of Cluster 1 have a greater influence on Target_Point. We can further assign weights to Cluster 1 and Cluster 2 based on the values ​​of Dist_Cluster_1 and Dist_Cluster_2, thereby quantifying the influence factor of the target layer on the soil characteristics.

[0054] In an embodiment of the present application, in order to determine the importance of each depth layer to the soil characteristics, soil parameter data containing multiple depth layers is collected, and each data point contains the soil parameter value of the target layer and the corresponding parameter values ​​of each reference. The data point can be expressed as (X_Target, Y_i_Reference), wherein X_Target represents the soil parameter value of the target layer, and Y_i_Reference represents the soil parameter value of the i-th reference layer. Then, a three-dimensional data space is constructed with the soil parameter value X_Target of the target layer as the horizontal axis, the soil parameter value Y_i_Reference of the i-th reference layer as the vertical axis, and the depth layer index (or depth value) of the data point as the third dimension. In this three-dimensional space, each data point can be expressed as a coordinate point (X_Target, Y_i_Reference, Depth_Index). In the three-dimensional data space of the target layer and each reference layer, all data points with the same horizontal coordinate value X_Target are selected, and these data points are projected onto a two-dimensional plane with Yi_i_Reference as the vertical axis and the data point index (or some kind of identifier) ​​as the horizontal axis to construct a two-dimensional data space. Each point in this two-dimensional space represents the soil parameter value of different depth layers (reference layers) under the same soil parameter value of the target layer. In each two-dimensional data space, the number of data points with the same or similar X_Target value is counted. These numbers can reflect the distribution of soil parameter values ​​of each reference layer under different soil parameter values ​​of the target layer. According to K-means clustering, distance difference method, etc., the influence factor of each depth layer (including the target layer and the reference layer) on the soil characteristics is calculated. Combined with the influence factor of each depth layer and the number of data points in the two-dimensional data space of the target layer and each reference layer, the importance of the target layer to the soil characteristics is evaluated. Specifically, the ratio between the influence factor of the target layer and the influence factor of the reference layer, as well as the ratio between the number of data points in the target layer and the number of data points in the reference layer can be calculated. These ratios can be used to quantify the relative importance of the target layer relative to the reference layer in the formation of soil properties.

[0055] In order to illustrate the determination process of the optimal clustering distance through data, a set of data points is assumed, each of which contains soil parameter values ​​of different depth layers and corresponding soil characteristic values, as shown in the following table:

[0056] Data point index D1 (depth layer 1 soil parameters) D2 (depth layer 2 soil parameters) D3 (depth layer 3 soil parameters) C (soil characteristic value) 1 15 20 25 7 2 16 19 24 8 3 14 21 26 6 4 17 18 23 9 5 13 22 27 5

[0057] Then, in the plane coordinate system, a distance measurement method is selected to calculate the initial cluster distance between any two data points. In the plane coordinate system of each depth layer, the Euclidean distance between any two data points is calculated, which will generate a set of initial cluster distance matrices corresponding to the D1, D2 and D3 depth layers respectively. The initial cluster distance is weighted and summed using the importance weight of the depth layer to obtain the weighted cluster distance between any two data points. The weighted cluster distance calculation formula is: D_weighted = w1D1 + w2D2 + w3*D3, where w1, w2, and w3 are the importance weights of the D1, D2, and D3 depth layers respectively, and D1, D2, and D3 are the initial cluster distances of the corresponding depth layers respectively. The weighted cluster distance is the preferred cluster distance, which takes into account the soil parameter values ​​of different depth layers and the importance of the depth layer to the soil characteristics.

[0058] In this application, in order to illustrate the process of determining the neighborhood radius and the minimum number of samples by the preferred clustering distance through data, a set of data points after the preferred clustering distance calculation, the preferred clustering distance between the data points is known, as shown in the following table:

[0059] Data Points A B C D A 0 2.5 3.2 4.1 B 2.5 0 1.8 2.9 C 3.2 1.8 0 2.2 D 4.1 2.9 2.2 0

[0060] It should be noted that the distance matrix is ​​symmetric and the distance from each data point to itself is 0.

[0061] Add up all the non-zero distances in the above matrix, and then divide by the total number of distances (that is, 6 pairs of distances between 4 points, but each pair of distances is calculated twice, so it must be divided by 2), mean = (2.5 + 3.2 + 4.1 + 2.9 + 1.8 +2.2) / 6 = 2.7833 (rounded to four decimal places, the actual calculation may be different according to the accuracy requirements), so we can set the neighborhood radius to 2.7833 (or rounded according to actual needs). For each data point, we find all other data points whose preferred clustering distance is less than the neighborhood radius. These points constitute the neighborhood data points of the data point. Add the number of neighborhood data points of all data points and divide it by the total number of data points to get the mean. Round this mean up to get the minimum number of samples. According to the neighborhood radius of 2.7833, we can determine the neighborhood data points of each data point:

[0062] A’s neighboring data points: B, C (because 2.5 and 3.2 are less than 2.7833);

[0063] B’s neighboring data points: A, C (because 2.5 and 1.8 are less than 2.7833);

[0064] C’s neighborhood data points: A, B (because 3.2 and 1.8 are less than 2.7833);

[0065] Neighborhood data points of D: None (because all distances are greater than 2.7833);

[0066] Count the number of neighborhood data points: A has 2, B has 2, C has 2, and D has 0.

[0067] Calculate the mean and round up: Mean = (2 + 2 + 2 + 0) / 4 = 1.5, which is rounded up to 2. Therefore, the minimum number of samples is set to 2.

[0068] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is limited by the attached embodiments and their equivalents.

Claims

1. A deep soil sampling method based on geological exploration, characterized in that: The method comprises the following steps: From the geological exploration area, obtain a dimensional sequence of soil quality information for each depth layer, wherein each depth layer corresponds to a soil quality parameter value; Based on the differences between soil parameters at each depth layer, the influencing factors of each depth layer on soil properties are obtained; Based on the differences among all depth layers and the influencing factors of each depth layer on soil properties, the weight of each depth layer on soil properties is obtained; Based on the weight of each depth layer on soil properties, the initial clustering distance between different depth layers is adjusted to obtain the optimal clustering distance; Based on the priority clustering distance, the neighborhood radius and the minimum number of samples are obtained respectively, and the density clustering algorithm is used to perform clustering operations on different depth layers to obtain the clustering results; Based on the clustering results, a soil property prediction model is constructed, and the soil properties of different depth layers are predicted using the model.

2. A deep soil sampling method based on geological exploration according to claim 1, characterized in that: The influencing factors of each depth layer on soil properties include the following steps: Record each depth layer as a data point; Among all the depth layers, any one depth layer is recorded as the target layer, and the other depth layers except the target layer are recorded as reference layers; Use the elbow method to operate on the i-th reference layer to obtain the optimal number of clusters for the i-th reference layer; Based on the optimal number of clusters in the i-th reference layer, the K-means clustering algorithm is used to cluster all data values ​​in the i-th reference layer to obtain several clusters; When the data values ​​in all reference layers corresponding to the ith data point are in the clusters corresponding to all references in the ith cluster combination, the ith data point is recorded as the mapping data point of the ith cluster combination.

3. The deep soil sampling method based on geological exploration according to claim 1 is characterized in that: The step of selecting a plurality of main mapping data points from all mapping data points based on each cluster combination; According to the differences between the soil parameter values ​​corresponding to the main mapping data points of each cluster combination, the influencing factors of the target layer on the soil characteristics are obtained.

4. The deep soil sampling method based on geological exploration according to claim 1 is characterized in that: The importance of each depth layer to soil properties includes the following steps: The soil parameter value of the target layer corresponding to the data point is taken as the horizontal axis, and the soil parameter value of the i-th reference layer corresponding to the data point is taken as the vertical axis, and a three-dimensional data space of the target layer and the i-th reference layer corresponding to all data points is constructed; In the three-dimensional data space of the target layer and each reference layer, a two-dimensional data space is constructed with all data points having the same horizontal coordinate value; Based on the influencing factor of each depth layer on soil properties and the number of data points in the two-dimensional data space divided by the three-dimensional data space of the target layer and each reference layer, the importance of the target layer to the soil properties is obtained.

5. The deep soil sampling method based on geological exploration according to claim 1 is characterized in that: The determination of the preferred clustering distance comprises the following steps: With the soil parameter value of each depth layer corresponding to the data point as the vertical axis and the soil characteristic value corresponding to the data point as the horizontal axis, a plane coordinate system for each depth layer corresponding to all data points is constructed; Based on the importance of each depth layer to soil properties and the initial cluster distance of any two data points on the plane coordinate system of each depth layer, the cluster distance between any two data points is obtained.

6. A deep soil sampling method based on geological exploration according to claim 1, characterized in that: The preferred clustering distance determines the neighborhood radius and the minimum number of samples, comprising the following steps: The mean of the preferred clustering distances between any two data points is recorded as the neighborhood radius; All other data points whose preferred clustering distance to the nth data point is less than the neighborhood radius are recorded as neighborhood data points of the nth data point; Count the number of neighborhood data points of each data point, round up the mean of the number of neighborhood data points of all data points, and record it as the minimum sample number.

Images